New detrital zircon U-Pb data from the Farewell terrane of interior Alaska illuminate its early provenance evolution and connections with other Alaskan terranes. Five samples come from Neoproterozoic units in the central Farewell terrane. Basal “ferruginous beds” and the overlying Windy Fork Formation have prominent detrital zircon age populations between 2000 and 1800 Ma, with the Windy Fork Formation also having major age peaks between 700 and 600 Ma. Younger (Lone Formation) samples yield grains mainly between 750 and 550 Ma, with fewer older Proterozoic grains. Eleven samples come from deep-water early Paleozoic rocks (southeastern Farewell terrane). Ordovician sandstone (Post River Formation) has a major age population at ca. 490 Ma and subordinate 785–550 Ma populations that overlap age peaks in the Lone Formation. Turbidites in the overlying Terra Cotta Mountains Sandstone (Silurian) yield distinctly different spectra, with major ca. 450–420 Ma age populations and numerous grains between 2000 and 900 Ma. Devonian Barren Ridge Limestone samples have spectra like those of the Terra Cotta Mountains Sandstone, plus some Early Devonian grains. The Silurian shift in detrital zircon age spectra coincides with a major influx of siliciclastic sediment suggestive of a tectonic (collisional?) event involving the Farewell terrane. Neoproterozoic through Devonian successions in the Arctic Alaska–Chukotka and Alexander terranes show a similar up-section shift in detrital zircon spectra, supporting links between these terranes and the Farewell terrane during the early Paleozoic. Detrital zircon ages from the White Mountains and Livengood terranes, adjacent to the northern Farewell terrane, include major early Paleozoic populations that overlap those seen in partly coeval Farewell strata.

The Farewell terrane is a continental fragment about the size of Switzerland in the interior of Alaska (Fig. 1) that includes strata of Proterozoic through Mesozoic age (Patton et al., 1994; Decker et al., 1994; Bradley et al., 2014). The Farewell terrane is one of several Alaskan terranes considered by many authors to be exotic to Laurentia and of inferred Arctic affinity (e.g., Colpron and Nelson, 2011, and references therein; Fig. 2). Lower Paleozoic rocks in the Farewell terrane have well-documented faunal and lithologic similarities with coeval strata in the central part of the Arctic Alaska–Chukotka composite terrane (Seward Peninsula, western and central Brooks Range; Figs. 1 and 2)—in particular, a highly distinctive early Paleozoic biota that includes Laurentian, Siberian, and some Baltic endemic forms (Dumoulin et al., 2002, 2012, 2014a; Blodgett et al., 2002; Rasmussen et al., 2012). Paleontologic and lithologic ties between the Farewell and Alexander terranes (Figs. 1 and 2), particularly during the Silurian, have also been established (Blodgett et al., 2002, 2010; Antoshkina and Soja, 2006, 2016; Soja, 2008). However, a detailed history of the Farewell terrane has not yet been delineated.

In this paper, we use new and previously published lithologic, paleontologic, and detrital zircon U-Pb data (Table 1) to define the Proterozoic through Devonian provenance evolution of the Farewell terrane. We then compare these findings to detrital zircon age and provenance patterns from other Alaskan terranes of Arctic affinity (Fig. 2). Detrital zircon data provide useful information that bears on terrane definitions and aids in interpreting terrane associations and histories within the complex tectonic collage of Alaska. Our data suggest modifications to previously proposed paleogeographic positions of the Farewell terrane (Colpron and Nelson, 2011; Beranek et al., 2013a) during the early Paleozoic.

The Farewell terrane (Decker et al., 1994; Bundtzen et al., 1997; Bradley et al., 2003) consists of a Proterozoic basement complex overlain by younger Proterozoic through Mesozoic strata that have been divided into several sequences or subterranes (Fig. 3). The older part of this stratigraphy includes a carbonate platform, the Nixon Fork subterrane, and its deep-water equivalent, the Dillinger subterrane. Strata of these two successions locally interfinger and grade into each other, and they are overlain by Devonian through Mesozoic strata of the Mystic subterrane.

The White Mountains and Livengood terranes (Fig. 1), as defined in Silberling et al. (1994), have lithologic and faunal features that suggest ties to the Farewell terrane (Blodgett et al., 2002; Dumoulin et al., 2014a). Lower Paleozoic rocks in the White Mountains terrane comprise a mafic volcanic succession (Ordovician Fossil Creek Volcanics) overlain by shallow-water carbonate strata (Silurian–Devonian Tolovana Limestone; Weber et al., 1992). Partly coeval rocks in the Livengood terrane include the Ordovician Livengood Dome Chert (Weber et al., 1992).

In this paper, we synthesize previously published detrital zircon age data with new lithologic information from Neoproterozoic strata of the Nixon Fork subterrane at Lone Mountain in the central McGrath quadrangle (Figs. 3 and 4). We then present new lithologic, faunal, and detrital zircon age data from Ordovician through Devonian rocks of the Dillinger subterrane in the eastern McGrath and Lime Hills quadrangles (Figs. 3 and 4). We also present new detrital zircon ages from lower Paleozoic strata in the Livengood and White Mountains terranes (Livengood quadrangle; Fig. 1).

Petrographic descriptions are based on examination of 180 samples in thin section, including all samples analyzed for detrital zircons (Table 1). Some thin sections were stained to facilitate identification of potassium and plagioclase feldspar. Chronostratigraphic correlations follow Walker et al. (2012).

Detrital zircon U-Pb age determinations were done by Apatite to Zircon (now Geosep Services) using laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) techniques at Washington State University. Supplemental Files1 include a detailed description of analytical methods, complete data tables, including U-Pb concordia diagrams for all samples (Table S1), and a summary of calculated age populations for each sample (Table S2). Raw, uninterpreted U-Pb data were also published as a U.S. Geological Survey (USGS) Data Release (Dumoulin et al., 2018). We calculated concordia ages (including decay constant errors) for each analysis and used this value as the preferred age of the associated detrital zircon grain, instead of using either the 206Pb/238U or 207Pb/206Pb age (Ludwig, 1998; Nemchin and Cawood, 2005). The concordia age makes optimum use of both decay schemes and obviates the need to choose an arbitrary age threshold for selecting the 206Pb/238U or 207Pb/206Pb age as the “preferred age” for an individual grain (Ludwig, 1998). Additionally, the concordia age calculation gives probability of concordance (POC) for each analysis, which provides a useful means of assessing concordance for all grains regardless of age. After calculating the concordia age and associated statistics for each analysis, we screened the data for uncertainty and POC. Analyses with greater than 10% age uncertainty (at 1σ) were excluded or “filtered” from plots and statistical treatments. Grains with POC <0.1 were also excluded unless the grain was older than 1000 Ma and had a calculated concordance (comparison of 206Pb/238U and 207Pb/206Pb ages) between 80% and 105%. Data that were excluded are reported in Table S1 (see footnote 1) but are highlighted to indicate that they were not considered further. We also calculated concordia ages for data from five samples previously published by Bradley et al. (2014) and screened these data using the same criteria described here to ensure consistency throughout the entire sample set considered herein. Age probability diagrams were generated using the kernel density estimation (KDE) method of Vermeesch (2012), and all KDE diagrams were generated using adaptive kernel density estimation, wherein the bandwidth varies depending on data density (Vermeesch, 2012). Histogram bins, where shown, generally represent ∼25 m.y. Age populations were identified by using the AgePick macro for Microsoft Excel (Gehrels, 2009), and results for all samples are shown in Table S2 (see footnote 1). The maximum depositional age for each sample was determined using the youngest age population made up of three or more grains with ages that overlapped within 2σ uncertainty and defined a distinct population with a mean square of weighted deviates (MSWD) of 2 or less. The maximum depositional age as determined from detrital zircon data for each sample is labeled with 2σ uncertainty and the number of grains contributing to the population in the KDE plots. Other notable age populations determined using the AgePick macro are also labeled in the KDE plots, where applicable, and discussed in the following text.

After screening all the data, we used the DZstats tool (version 2.2) of Saylor and Sundell (2016) to plot the cumulative distribution function for different sample sets and to calculate a variety of comparative metrics, including Kolmogorov-Smirnoff (K-S) and Kuiper tests and cross-correlation, likeness, and similarity coefficients of probability density plots (PDPs) for all samples. The results of the statistical comparisons for all samples are shown in the Supplemental File (Table S3), and select results are discussed in the text here. We also applied multidimensional scaling (MDS) using the R code of Vermeesch et al. (2016) as another means of assessing the similarity between our samples and other published data. The MDS plots of the U-Pb data sets use the K-S effect size as a dissimilarity measure. MDS is a statistical technique that uses pairwise calculated dissimilarities between samples to produce a map of points on which more similar samples cluster closely together and more dissimilar samples plot farther apart (Vermeesch, 2013; see following).

Stratigraphy and Lithologies

Strata at Lone Mountain represent an isolated outlier of the Nixon Fork carbonate platform, separated by large tracts of Quaternary sediment from more continuous exposures of the Nixon Fork carbonate platform ∼25 km to the south and 50 km to the north (Fig. 4). At Lone Mountain (Figs. 5 and 6), unnamed, chiefly shallow-water carbonate rocks of Cambrian and Ordovician age overlie a late(?) Neoproterozoic sedimentary succession that was described in detail by Babcock et al. (1994). To the south and southwest, at White Mountain and in the Sleetmute and Taylor Mountains quadrangles (Fig. 4), the Nixon Fork subterrane consists of Cambrian through Devonian carbonate rocks and limited exposures of Neoproterozoic(?) strata similar to those at Lone Mountain (Wilson et al., 1998; LePain et al., 2000). To the northeast, in the Medfra quadrangle (Fig. 4), Ordovician–Devonian carbonate rocks of the Nixon Fork overlie a basement of amphibolite- to greenschist-facies meta-igneous and metasedimentary rocks. The oldest reliably dated protoliths in this basement, based on U-Pb zircon ages, are early Neoproterozoic (Tonian) meta-rhyolites with ages of ca. 979 and 921 Ma, and ca. 852–850 Ma granitic orthogneisses (McClelland et al., 1999; Bradley et al., 2003, 2014).

Neoproterozoic strata at Lone Mountain are not metamorphosed and comprise two carbonate units, the Khuchaynik and Big River dolostones, intercalated with three siliciclastic units (Figs. 57). The carbonate units consist of limestones and dolostones with well-preserved microtextures that include algal features and a variety of coated grains (Fig. 5B). A Neoproterozoic age for these strata is supported by an absence of metazoan fossils (Babcock et al., 1994), the presence of distinctive large coated grains found elsewhere in Arctic Alaska in Neoproterozoic strata (e.g., Dumoulin, 1988; Clough and Goldhammer, 2000), and detrital zircon data, discussed later herein. Overlying strata contain middle Cambrian trilobites (Babcock et al., 1994).

The basal unit at Lone Mountain (Fig. 5), informally called the “ferruginous beds” (Babcock et al., 1994), is made up of silty shale, fine-grained sandstone, and sandy conglomerate. Sand and pebbles (4 cm maximum diameter) are well rounded, have thin ferruginous coatings, and are cemented by silica (Fig. 7A). Clasts are mainly monocrystalline quartz (sand), and polycrystalline quartz (pebbles), with 5%–10% potassium feldspar grains and rare pebbles of quartz-potassium feldspar sandstone (Figs. 7A and 7B).

Stratigraphically higher siliciclastic units—the Windy Fork and Lone Formations—are composed of orange- to pink- or rusty-weathering siltstone and sandstone made up mostly of subangular monocrystalline quartz, rounded micritic clasts, coated grains, and rhombic dolomite detrital grains and cement (Figs. 5A, 5C, and 7C–7G). Both units have subordinate dolostone interbeds. Windy Fork sandstone samples are very fine to fine grained with ≤40% carbonate and minor amounts of chert grains and white mica (Figs. 7C and 7D). Lone Formation sandstone is very fine to coarse grained and contains up to 90% carbonate, chiefly dolomite; angular and less-common rounded grains of monocrystalline quartz, typically outsized, occur in most samples (Figs. 7E–7G).

Sedimentary features suggest a supratidal to shallow subtidal setting for the entire succession (Babcock et al., 1994). These features include fenestral fabric and oolitic-pisoidal grainstone (Fig. 5B) in the carbonate units, and planar cross-laminae (Fig. 7E) and symmetrical ripple marks (Babcock et al., 1994) in the siliciclastic strata.

Detrital Zircon Data

Detrital zircon data from the Lone Mountain succession were first published by Bradley et al. (2014) and have been recalculated, screened, and plotted by the methods described for this study (Fig. 8). Samples (Figs. 5 and 6; Table 1) consist of one each from the ferruginous beds (L1) and the Windy Fork Formation (L2), and three from the Lone Formation (L3–L5). Four samples come from the ridge where Babcock et al. (1994) measured and named the Neoproterozoic units (Fig. 6, L1–L4); a fifth sample, from a ridge 2 km to the east (Fig. 6, L5), was originally mapped as Cretaceous Kuskokwim Group (Bundtzen inWilson et al., 1998), but it is compositionally similar to the Lone Formation, and we include it in that unit.

Detrital zircon samples show a change in age spectra upward through the section, with increasing proportions of Neoproterozoic grains and variable distributions of older Precambrian grains (Fig. 8). The ferruginous beds sample has a broad distribution of Paleoproterozoic detrital zircon ages between ca. 2000 and 1800 Ma, with a maximum at ca. 1864 Ma and a subsidiary age population at ca. 1943 Ma (Fig. 8; Table S1 [footnote 1]). The overlying Windy Fork Formation has a broad distribution of Paleoproterozoic grains between ca. 2050 and 1850 Ma, with a maximum at ca. 1991 Ma, and a younger broad distribution of Neoproterozoic and earliest Cambrian grains between ca. 750 and 539 Ma, with a maximum of ca. 613 Ma and another population at ca. 685 Ma. Three samples from the Lone Formation yielded mainly Neoproterozoic grains (62%–79% of total grains among three samples) with a subordinate Paleoproterozoic component (14%–26%). All three Lone Formation samples have prominent Neoproterozoic age distributions between ca. 750 and 550 Ma, similar to those seen in the Windy Fork sample. Mesoproterozoic grains (between 1600 and 1000 Ma) are rare throughout the Lone Mountain succession, making up 5% or less of the total population in most samples. Youngest zircon age populations get younger upward through the section, with a calculated maximum depositional age of 1372 ± 17 Ma in the ferruginous beds, 613 ± 8 Ma in the Windy Fork Formation, and 625 ± 9–570 ± 7 Ma in the Lone Formation.

There are some general similarities in age populations among these samples, but quantitative and graphical comparisons show that only the three Lone Formation samples are statistically similar when all data are compared (Fig. 8; Table S3 [footnote 1]). The cumulative distribution function (Fig. 8B) shows that there is overlap between the youngest 25% of the age spectra for the Windy Fork and Lone Formations, but age distributions and grain proportions of the rest of the spectra are notably different. The ferruginous beds are distinct from all of the overlying strata because of the more prominent Paleoproterozoic age populations and a near-complete absence of Neoproterozoic grains.

Stratigraphy and Lithologies

Cambrian and younger strata that overlie the Neoproterozoic succession at Lone Mountain, like coeval rocks elsewhere in the Nixon Fork subterrane, are mainly carbonate rocks with little siliciclastic material and are not prospective for detrital zircon sampling. Deeper-water, siliciclastic-rich strata of the correlative Dillinger subterrane, however, are a more promising target and were sampled at 11 localities in the McGrath and Lime Hills quadrangles southeast of Lone Mountain (Fig. 9).

The Dillinger succession (Fig. 10) begins with upper Cambrian to lowest Ordovician parallel- to cross-laminated silty limestone of the Lyman Hills Formation (Fig. 10F); some intervals are graded and ripple-laminated and may represent partial Bouma sequences (Bundtzen et al., 1994). This unit is “essentially equivalent” to map unit OCls in the shallow-water succession at Lone Mountain, and thus it is an important link between the Nixon Fork and Dillinger subterranes (T.K. Bundtzen, 1997, written commun.).

Dark shale and argillite dominate the Post River Formation (400 m thick; Fig. 10E), which is well dated by graptolites from 14 zones that span 45 m.y. of Early Ordovician through early Silurian time (Churkin and Carter, 1996). Fine-grained limestone and radiolarian chert form local interbeds. Rare quartz-rich siltstone and very fine-grained sandstone beds (Fig. 11A), typically ≤5 cm thick but rarely as much as 15 cm, are found throughout the unit; fine grain size and bottom features such as flute, load, and groove casts indicate a distal turbidite origin for these beds.

The overlying Terra Cotta Mountains Sandstone (Churkin and Carter, 1996), a thick (>700 m) sequence of fine- to coarse-grained turbidites (Figs. 10C and 10D), marks a major increase in the grain size and amount of siliciclastic material deposited in the Dillinger basin. Siltstone, sandstone, and lesser conglomerate contain varied amounts of carbonate, quartz, feldspar, mica, and metamorphic and sedimentary lithic grains (Figs. 11B–11D); lithic grains make up 20%–80% of clasts in our samples. Carbonate makes up >50% of most samples but is locally absent. It occurs chiefly as calcite cement and micritic or peloidal clasts; a few clasts have compositions, e.g., oncoid grainstone (Fig. 11C), that indicate a shallow-water, carbonate-platform source. Detrital mica (muscovite, chlorite, and biotite) is a distinctive minor component of virtually all samples (Fig. 11D). Sedimentary features in the Terra Cotta Mountains Sandstone include flute casts, ripples, graded beds, and partial to complete Bouma sequences. Calcitized radiolarians are common in subordinate limestone beds. Graptolites in shale layers indicate a middle Silurian (Wenlock) age for the lower part of the unit (Churkin and Carter, 1996); age constraints from the overlying Barren Ridge Limestone (see following) imply that the youngest beds are no younger than late Silurian. Sediment accumulation rates, calculated using published unit thicknesses and age data and uncorrected for compaction, are 6–8 times higher in the Terra Cotta Mountains Sandstone than in the Post River Formation.

The uppermost unit in the Dillinger succession is the Barren Ridge Limestone (Churkin and Carter, 1996), which consists mainly of fine-grained limestone—locally silty—with subordinate coarser beds of very fine- to fine-grained sandstone and rare conglomerate (Figs. 10A and 10B). Sedimentary features, including graded bedding and partial Bouma sequences (Bundtzen et al., 1997), suggest a turbidite origin for this unit. Composition of coarser beds is generally similar to that of sandstone and conglomerate in the Terra Cotta Mountains Sandstone, with a mix of carbonate clasts, quartz, feldspar, and metamorphic and sedimentary lithic grains (Figs. 11E and 11F). Overall, however, Barren Ridge Limestone samples are more calcareous and less micaceous and include locally notable volcanic lithic grains (Fig. 11F), which are rare in our Terra Cotta Mountains Sandstone samples. The Barren Ridge Limestone has not yielded age-diagnostic fossils in its type area in the southeastern McGrath quadrangle, but strata considered to be correlative in the Lime Hills D-4 quadrangle contain Early Devonian (Lochkovian to Pragian) conodonts, as well as a predominantly pelagic fauna of dacryoconarid tentaculitids and orthoconic nautiloids (Blodgett and Gilbert, 1992).

Detrital Zircon Data

Detrital zircon data from Dillinger strata, like those from the Neoproterozoic succession at Lone Mountain, show an up-section evolution with younging maximum depositional ages (Figs. 12 and 13). Two samples of dolomitic siltstone from the Lyman Hills Formation contained few zircons, but a bed of very fine-grained sandstone in the Post River Formation was productive (Table 1; Figs. 9, 10, and 12). The Post River Formation sample is mainly monocrystalline quartz with subordinate dolomite and plagioclase feldspar (Fig. 11A); interbedded shales have a rich graptolite fauna of late Middle to early Late Ordovician age (late Darriwilian–early Sandbian; S. Finney, 2013, written commun.; Table S4 [footnote 1]). The detrital zircon spectrum has a youngest population, and thus a maximum depositional age, of 491 ± 7 Ma (latest Cambrian) and a prominent distribution of Neoproterozoic ages, the youngest of which overlap age populations in the Lone Formation (Fig. 12). Mesoproterozoic grains are rare (5%). The age spectrum of the Post River Formation is not statistically similar to spectra of any other units within this study, although the MDS plot (Fig. 12C) shows that the Post River Formation is more similar to the Lone Formation than to any other unit.

Our five samples of Terra Cotta Mountains Sandstone came from the McGrath and Lime Hills quadrangles (Table 1; Figs. 9, 10, and 12) and include one from the type area of the formation (Churkin and Carter, 1996) and two from beds just to the west that overlie a layer with Wenlock graptolites (loc. 10 inBundtzen et al., 1987). Samples range from very fine-grained to pebbly sandstone, contain 20%–40% noncarbonate grains, and yielded a consistent set of detrital zircon spectra (Fig. 12). All produced abundant Late Ordovician–Silurian grains (450–420 Ma) and numerous grains between 2000 and 900 Ma—including a notable component of Mesoproterozoic grains (1600–1000 Ma) that make up 35% of the composite sample. Subordinate populations at ca. 495–470 Ma, which overlap the ca. 491 Ma population in the Post River Formation sample, are present in all of the Terra Cotta Mountains Sandstone samples (Fig. 12A; Table S2 [footnote 1]). Maximum depositional ages calculated from detrital zircon data range from 430 ± 5 Ma to 423 ± 5 Ma, consistent with the middle to late Silurian age indicated for this unit by fossil data. The MDS plot (Fig. 12C) shows that all of the Terra Cotta Mountains Sandstone samples cluster together around the composite data set. Quantitative metrics show some variation in the similarity of the five samples (Table S3 [footnote 1]). Cross-correlation coefficients of PDPs range from 0.16 to 0.53 for all intersample comparisons, with sample 11AD22A producing the widest range of values. Similarity and likeness coefficients for intersample comparisons range from 0.77 to 0.97 and from 0.36 to 0.62, respectively (Table S3).

Five samples of relatively siliciclastic-rich, very fine-grained sandstone to conglomerate from the Barren Ridge Limestone (Table 1; Fig. 10) produced an internally consistent set of detrital zircon spectra much like those from the Terra Cotta Mountains Sandstone (Fig. 13), but with all having Early Devonian youngest grain populations and maximum depositional ages. All samples were from the McGrath quadrangle (Fig. 9) and include four from the type area of the Barren Ridge Limestone (Churkin and Carter, 1996). Three spectra have subordinate late Cambrian–Early Ordovician age populations like those from the Terra Cotta Mountains Sandstone. Youngest age populations are Lochkovian to early Emsian, with calculated ages of 416 ± 5 Ma to 404 ± 5 Ma (Fig. 13; Table S2 [footnote 1]) that match well with the sparse fossil data from this unit. Multiple statistical comparisons show significant similarity among Barren Ridge Limestone samples, although there is slightly more disparity between samples than observed within the Terra Cotta Mountains Sandstone (Table S3). There is also significant similarity between many of the Barren Ridge Limestone and Terra Cotta Mountains Sandstone samples (Table S3), suggesting some degree of common provenance for the two units. Similarity among Barren Ridge Limestone samples and the Terra Cotta Mountains Sandstone composite spectrum is illustrated in the MDS plot in Figure 13C, wherein the points generally cluster together with respect to other older successions. It is worth noting that one Barren Ridge Limestone sample (07ADw715A) is more similar to the Terra Cotta Mountains Sandstone composite spectrum than to the other Barren Ridge Limestone samples, likely because of the higher proportion of Neoproterozoic and Mesoproterozoic grains (Fig. 13).

The Livengood and White Mountains terranes of interior Alaska (Figs. 1 and 14) contain lower Paleozoic strata that have faunal and lithologic similarities to coeval rocks of the Farewell terrane (Blodgett et al., 2002; Dumoulin et al., 2014a). New detrital zircon data from both of these terranes allow us to test these correlations (Table 1; Fig. 15).

The most widely distributed unit in the Livengood terrane is the Ordovician Livengood Dome Chert, a succession of radiolarian chert and lesser mudrock, with rare siltstone, sandstone, limestone, and mafic volcanic rocks (Fig. 14; Chapman et al., 1980). Early to earliest Middle Ordovician conodonts and Late Ordovician (Ashgill) graptolites occur in this unit (Chapman et al., 1980; Weber et al., 1994; Dumoulin and Harris, 2012; Dumoulin et al., 2014a). Fine to medium-grained, volcaniclastic sandstone from the reference section produced a unimodal peak (and maximum depositional age) of 486 ± 5 Ma (latest Cambrian; Fig. 15). No older zircons were found in this sample.

Partly correlative, but lithologically distinct, lower Paleozoic strata occur in the White Mountains terrane (Figs. 1 and 14). Alkali basalt, agglomerate, and volcaniclastic conglomerate, with subordinate limestone and feldspathic sandstone, make up the Fossil Creek Volcanics, thought to have formed in an extensional setting along a continental margin (Wheeler et al., 1987; Weber et al., 1992). Abundant fossils have been interpreted to indicate an age of early Early Ordovician (Tremadoc) for the lower part of the unit and late Late Ordovician (Ashgill) for the upper beds (Blodgett et al., 1987; Weber et al., 1994); the age of magmatism is uncertain. These rocks are overlain by the Tolovana Limestone, which contains conodonts of early Silurian (early to middle Llandovery) age a few meters above the base and younger Silurian and Devonian fossils higher in the unit (Blodgett et al., 1987; Weber et al., 1994).

Our two detrital zircon samples came from section A of Blodgett et al. (1987) and Wheeler et al. (1987), where the contact between the Fossil Creek Volcanics and the Tolovana Limestone is marked by 6.6 m of feldspathic sandstone with limestone interbeds that contain Ashgill brachiopods and corals (Fig. 14B). Blodgett et al. (1987) and Wheeler et al. (1987) interpreted this interval as the uppermost part of the Fossil Creek Volcanics, but Oliver et al. (1975) considered it to be a distinct unit (“tuffaceous limestone” in their figure 10) beneath the Tolovana Limestone. The uppermost bed in the interval, apparently unfossiliferous and described by Blodgett et al. (1987, p. 54) as “iron-stained calcareous sandstone, with…a paleokarst surface,” was sampled (14AJJ207A; Figs. 1 and 15) and consists of fine- to medium-grained, calcite-cemented sandstone with abundant mafic volcanic lithic fragments, and lesser quartz and feldspar. It has a peak (and maximum depositional) age of 438 ± 5 Ma (early Silurian, i.e., Llandovery; Fig. 15A). A second detrital zircon sample was collected from sand-sized matrix in prominent volcaniclastic conglomerate beds ∼8 m beneath the base of the Tolovana Limestone. The matrix-supported conglomerate contains well-rounded clasts that are chiefly igneous and range from 1 to 15 cm in diameter. The sample produced a single dominant detrital zircon population with an age of 436 ± 5 Ma (early Silurian, i.e., Llandovery; Fig. 15A). Precambrian grains make up ∼20% of the two Fossil Creek Volcanics samples and include minor Paleoproterozoic (ca. 1811–1735 Ma) and Neoarchean (ca. 2516 Ma) populations (Table S2 [footnote 1]).

Our detrital zircon data indicate a Silurian maximum age for at least part of the Fossil Creek Volcanics and thus are in conflict with the previously reported Ordovician age that was based on fossils. One possible resolution of this conflict is that the Late Ordovician fossils in the upper part of the Fossil Creek Volcanics (Blodgett et al., 1987) were eroded and redeposited into younger (early Silurian?) strata. Additional sampling of the Fossil Creek Volcanics at multiple localities, and detailed study of the unit’s contacts with adjacent strata, may shed light on this dilemma.

Comparisons within the Farewell Terrane

Detrital zircon data from the base of the Lone Mountain succession have some similarities with data from other parts of the Farewell terrane, but results from higher beds in this area do not. Four quartzites—two from the basement complex in the Ruby quadrangle and two from Cambrian (or older?) strata in the Taylor Mountains quadrangle (Fig. 4)—produced age spectra that, like the ferruginous beds spectrum, are dominated by Paleoproterozoic detrital zircon (Figs. 16U and 16V; Bradley et al., 2007, 2014). The four quartzite samples have populations ranging from ca. 2050 Ma to 1978 Ma; the youngest significant age populations are all older than 800 Ma. Age spectra dominated by Neoproterozoic peaks, however, like those from the Lone Formation, have not been seen elsewhere in the Farewell terrane.

Our findings in the Dillinger subterrane compare well with preliminary results of comparable studies by other workers. Unpublished detrital zircon data from the Terra Cotta Mountains Sandstone, reported in Koroleski et al. (2012), Hampton et al. (2013), Koroleski and Hampton (2014), and Koroleski (2014), are similar to those we obtained from this unit; their samples came from three localities (one in the eastern McGrath quadrangle and two in the northwestern Lime Hills quadrangle; Koroleski, 2014). Two samples of Barren Ridge Limestone(?) in the western Talkeetna quadrangle yielded spectra more like those from the Terra Cotta Mountains Sandstone (Koroleski, 2014). Paleozoic and older detrital zircons from the Dillinger subterrane samples in these studies have both enriched and depleted epsilon Hf values (Hampton et al., 2013); Paleozoic grains from three samples of the Terra Cotta Mountains Sandstone have epsilon Hf values that range from –13.9 to +13 (Koroleski, 2014).

Provenance Evolution of the Farewell Terrane

For Proterozoic through Devonian strata of the Farewell terrane, our data indicate an evolution in the detrital zircon spectra that corresponds to changes in sandstone composition (Figs. 16O–16T). Quartz-rich sandstones of early(?) Neoproterozoic age are widely distributed within the Farewell terrane and include both recrystallized quartzites of the basement complex in the northeast (locations R1 and R2, Fig. 4) and unmetamorphosed sandstones at Lone Mountain (ferruginous beds, location L1, Fig. 6). Similar samples in the southwest (locations TM1 and TM2, Fig. 4) come from isolated outcrops of uncertain depositional age that are near definitively Cambrian strata but could be older. All of these rocks produced mostly Paleoproterozoic detrital zircons, largely ca. 2000–1800 Ma in age (Figs. 16T–16V).

Late Neoproterozoic sandstones at Lone Mountain consist mainly of quartz and carbonate clasts. The older Windy Fork Formation yielded both Paleoproterozoic and late Neoproterozoic detrital zircons (ca. 750–570 Ma), whereas samples from the younger Lone Formation produced mainly late Neoproterozoic grains (Figs. 16R and 16S). Similar-aged late Neoproterozoic zircons are abundant in a quartz-rich Ordovician sandstone from the Post River Formation, which also contains a prominent population of latest Cambrian–earliest Ordovician (ca. 490 Ma) grains (Fig. 16Q). Age populations between ca. 900 and 800 Ma are present in all spectra from the Windy Fork, Lone, and Post River Formations and could reflect input from Tonian igneous rocks of the Farewell basement complex (Bradley et al., 2014). The Post River Formation sample provides an important link between the Nixon Fork and Dillinger subterranes, because it suggests that rocks at the base of and underlying the Nixon Fork carbonate platform were eroded and redeposited into strata of the Dillinger basin.

As noted already, the Terra Cotta Mountains Sandstone marks a major shift in grain size and composition of the Farewell succession. The Terra Cotta Mountains Sandstone records an influx of carbonate-siliciclastic turbidites containing abundant sedimentary and metamorphic lithic grains that are absent from older units (Table 1; cf. Figs. 7A–7G, 11A, and 11B–11D), as well as a large component of latest Ordovician–Silurian (ca. 450–420 Ma) detrital zircons (Fig. 16P). Proterozoic zircons in these samples are mainly 2000–900 Ma, with more abundant Mesoproterozoic grains than are seen in older Farewell strata. Fossil and detrital zircon age data indicate that this sediment incursion began in the middle Silurian (Wenlock) and had largely ended by latest Silurian time. Early Devonian deposits of the Barren Ridge Limestone are carbonate-dominated mass flows and turbidites with a greatly reduced siliciclastic component. Composition and detrital zircon spectra of the subordinate siliciclastic-rich layers in this unit are quite similar to those of the Terra Cotta Mountains Sandstone (Fig. 16O), but they include Early Devonian grains and a smaller proportion of Precambrian grains.

The sediment influx recorded by the Terra Cotta Mountains Sandstone is suggestive of an as-yet-unknown tectonic event involving the Farewell terrane. Deposition of the Terra Cotta Mountains Sandstone coincided with prolonged Silurian drowning of the Nixon Fork carbonate platform (Dumoulin et al., 2002), leading Bradley (2008, p. 2 of the supplementary data) to interpret these turbidites as orogenically derived flysch that accumulated in the foredeep of an arc–passive-margin collision zone; continent-continent collision is another plausible tectonic scenario (Dumoulin et al., 2012). Possible sediment sources for the Terra Cotta Mountains Sandstone are discussed further in the following.

Detrital zircon age patterns reflect the tectonic settings of the sedimentary strata that produce them (Cawood et al., 2012), and our detrital zircon spectra yield insights into the tectonic history of the Farewell terrane (Fig. 16X). Although this terrane is widely described as a deformed but coherent continental-margin sequence (e.g., Decker et al., 1994), details of its origin and history remain uncertain. Bradley (2008, p. 2 of the supplementary data) suggested that at least part of the Nixon Fork carbonate platform represents a passive margin—most clearly, Ordovician strata deposited between ca. 485 Ma and 450 Ma. Strata of this age are >3 km thick and show an exponentially declining subsidence rate, consistent with a passive-margin setting (Dumoulin et al., 1998; Bradley, 2008). Details of the initiation and demise of this margin, however, remain problematic. Bradley (2008) suggested start and end dates of 545 Ma (early Cambrian) and 435 Ma (early Silurian), but deep-water strata as old as Late Ordovician interfinger with shallow-water strata in the northern Nixon Fork subterrane (Dumoulin et al., 2002, 2014a), suggesting incipient drowning of the platform by this time. Cumulative distribution function (CDF) plots for our detrital zircon data from the Farewell terrane (Fig. 16X) indicate mostly active margin settings on the diagram from Cawood et al. (2012). In particular, data from both the Post River Formation and the Terra Cotta Mountains Sandstone denote a collisional margin setting. The age of the Post River Formation sample is ca. 460–456 Ma, based on graptolites (Table S4 [footnote 1]; absolute age data from Walker et al., 2012), supporting the hypothesis that the collisional demise of the Nixon Fork passive margin may have begun in the Late Ordovician.

Livengood and White Mountains Terranes

New detrital zircon data from the Livengood and White Mountains terranes have some similarities to data discussed here from the Farewell terrane (Figs. 15 and 16). The late Cambrian peak detrital zircon population from the Livengood Dome Chert (ca. 486 Ma; Fig. 16N) is close in age to the younger of the two predominant age populations in the Post River Formation (ca. 490 Ma; Fig. 16Q). The dominant Silurian populations in the Fossil Creek Volcanics (438 ± 5 and 435 ± 5 Ma; Fig. 15) are similar to prominent age populations in Silurian and Devonian strata in the Dillinger subterrane (Figs. 13 and 14). These data suggest that common sources could have provided Cambrian detrital zircons to the Farewell and Livengood terranes, and Silurian detrital zircons to the Farewell and White Mountains terranes. The relatively few Precambrian detrital zircons in the White Mountains samples do not indicate a single, distinctive source. CDF data (Fig. 16X) plotted on the diagram from Cawood et al. (2012) suggest an active (convergent) margin setting for both the Livengood and White Mountains samples.

Comparison with the Arctic Alaska–Chukotka Terrane

Detrital zircon spectra from Paleozoic rocks of the central Arctic Alaska–Chukotka composite terrane (Figs. 1 and 2) show an up-section age progression that resembles the progression detailed here in the Farewell terrane—that is, a shift from spectra dominated by late Neoproterozoic grains to spectra with abundant Silurian and Mesoproterozoic grains. This is particularly evident in data from the Nome Complex on Seward Peninsula (Till et al., 2014a). Though deformed and metamorphosed to blueschist facies, the Nome Complex contains a mappable stratigraphy with age constraints provided by rare fossils and detrital zircons. An older rift basin of probable Ordovician age is overlain and underlain by carbonate strata, which are in turn succeeded by a siliciclastic succession of Devonian or younger age (Dumoulin et al., 2014a; Till et al., 2014b).

Ordovician rift deposits (mafic meta-turbidites, unit O_s) in the Nome Complex contain detrital zircons with ages mainly between 750 and 550 Ma (Fig. 16L; Till et al., 2014a). These ages overlap most of the detrital zircon ages from the Lone Formation and part of the spectra from the Windy Fork Formation in the Farewell terrane (Figs. 16R and 16S, respectively). Similar detrital zircon ages are found in Ordovician and older sedimentary rocks elsewhere in Arctic Alaska: unit O<t in the York Mountains of western Seward Peninsula (Fig. 16K; Fig. S1 [footnote 1]; Tables S1, S3, and S5 [footnote 1]; Amato et al., 2009; Dumoulin et al., 2014a) and in the central Brooks Range at Snowden Mountain (Moore, 2012). In addition, Lower Ordovician quartzose limestone (unit Oal) in the York Mountains (Fig. 16J; Table S5 [footnote 1]) yielded a detrital zircon age spectra with some similarities to that of the Post River Formation; both units contain latest Cambrian–Early Ordovician and late Neoproterozoic age populations (Dumoulin et al., 2014a; Fig. S1; Table S1 [footnote 1]). Statistical comparisons of these strata demonstrate the similarity of the Lone Formation with both the Ordovician strata (unit O_s) of the Nome Complex and unit O<t in the York Mountains, yielding PDP cross-plot correlation coefficients of 0.76 and 0.55, respectively (Table S3 [footnote 1]; Saylor and Sundell, 2016). Likeness and similarity coefficients are also higher for these comparisons (Table S3), and these units cluster together in the MDS plot in Figure 16W. Comparison of the York Mountains unit Oal with the Post River Formation produced a lower PDP cross-plot coefficient of 0.34 (Table S3; Saylor and Sundell, 2016) but somewhat higher coefficients of likeness (0.53) and similarity (0.79).

The younger—likely Devonian—part of the Nome Complex produced a distinctly different detrital zircon population (Figs. 16H and 16I; Till et al., 2014a) that resembles those from Silurian and Devonian Dillinger strata. A few samples produced chiefly Mesoproterozoic grains, largely between 1250 and 900 Ma in age (Fig. 16I; Mesoproterozoic theme samples of Till et al., 2014a). Most samples are dominated by early Paleozoic grains between 450 and 420 Ma, with a subordinate but notable component of Mesoproterozoic grains (Fig. 16H; Paleozoic theme samples of Till et al., 2014a). Samples with the Paleozoic detrital zircon theme (Fig. 16H) are statistically similar to the Silurian Terra Cotta Mountains Sandstone samples from the Farewell terrane (Fig. 16P), producing a PDP cross-correlation coefficient of 0.78 (Table S3 [footnote 1]; Saylor and Sundell, 2016). Predominant detrital zircon age populations in the Lower Devonian Barren Ridge Limestone are ca. 416 Ma (Fig. 16O), which is younger than the youngest age populations in the Nome Complex Devonian units. Again, similar detrital zircon age spectra are reported from possibly correlative samples elsewhere in Arctic Alaska, such as the Devonian–Lower Mississippian Endicott Group of the Brooks Range (Moore et al., 2004, 2007). Rocks on Wrangel Island, in the western part of the Arctic Alaska–Chukotka terrane, show an analogous age progression, with an older (Devonian–Carboniferous?) sample producing mainly 800–500 Ma detrital zircons, and a younger, Carboniferous sample dominated by 2000–900 Ma grains (Miller et al., 2010).

Comparison with the Alexander Terrane

The Alexander terrane in southeastern Alaska and western Canada (Figs. 1 and 2) is a Neoproterozoic–Jurassic crustal fragment (Beranek et al., 2013b) that includes lower Paleozoic strata with paleontologic links to the Farewell terrane, particularly in faunas of late Silurian–Middle Devonian age (Blodgett et al., 2002, 2010; Antoshkina and Soja, 2006, 2016; Soja, 2008). Recently published detrital zircon data from the Alexander terrane (Figs. 16A–16G; Beranek et al., 2013a, 2013b; Tochilin et al., 2014; White et al., 2015) show similarities and differences with data from the Farewell terrane. Quartzose sandstones from the upper Cambrian–Middle Ordovician Donjek assemblage (northern Alexander terrane) yielded predominant age populations ranging from 760 to 565 Ma (Fig. 16G; Beranek et al., 2013a), which overlap ages of the Neoproterozoic detrital zircon components in our Lone and Windy Fork Formation samples, but also contain more Mesoproterozoic grains (cf. Figs. 16G to 16R–16S). Statistical comparison of the Donjek quartzose sandstones and the Lone Formation produces a PDP cross-correlation coefficient of 0.58 (Table S3 [footnote 1]; Saylor and Sundell, 2016), suggesting that they are fairly similar. Statistical comparison of the Donjek sandstones and the Windy Fork Formation yields a much lower PDP cross-correlation coefficient of 0.26 (Table S3), likely because of the significant proportions of Paleoproterozoic and Neoarchean detrital zircon grains in the Windy Fork Formation.

Volcaniclastic sandstones from the Donjek assemblage are distinct from the quartzose sandstones in that they have a unimodal detrital zircon age population at 477 Ma (Fig. 16F; Beranek et al., 2013a). The peak age is slightly younger than the major peaks in the Post River Formation and Livengood Dome Chert samples (Figs. 16Q and 16N). However, overlap between the Donjek volcaniclastic sandstones and the Livengood Dome Chert produces a PDP cross-correlation coefficient of 0.86 (Table S3 [footnote 1]; Saylor and Sundell, 2016), indicating that they are statistically similar. Ordovician(?) quartzose sandstones from the Banks Island assemblage have a prominent age probability peak at 478 Ma (Fig. 16E; Tochilin et al., 2014) and are also statistically similar to the Donjek volcaniclastic sandstones and Livengood Dome Chert (PDP cross-correlation values of 0.91 and 0.77, respectively; Table S3 [footnote 1]; Saylor and Sundell, 2016). Thus, it is possible that all of these units were derived from the same or similar sources.

Lower Paleozoic strata elsewhere in the Alexander terrane, such as the Ordovician “Moira group” and the Ordovician–Silurian Descon Formation, have spectra with peak ages of ca. 460 Ma (Figs. 16D and 16C; Tochilin et al., 2014). Silurian–Lower Devonian strata from the Icefield assemblage (Fig. 16B; Beranek et al., 2013b), as well as age-equivalent and possibly correlative rocks from the Banks Island assemblage (Fig. 16A; Tochilin et al., 2014), contain detrital zircon age populations that are generally similar to those from coeval Dillinger subterrane units (Figs. 16O and 16P), but with fewer Mesoproterozoic grains; peak ages are between ca. 445 and 430 Ma. Statistical comparisons between the Silurian–Devonian components of the Icefield assemblage and Banks Island assemblage and the Terra Cotta Mountains Sandstone produce PDP cross-correlation coefficients of 0.56 and 0.58, respectively, suggesting that they are somewhat similar (Table S3 [footnote 1]; Saylor and Sundell, 2016). Comparison of these strata with the Fossil Creek Volcanics produced higher PDP cross-correlation coefficients of 0.72 and 0.67, respectively (Table S3).


U-Pb zircon data, in conjunction with paleontologic and lithologic evidence, illuminate the early history of the Farewell terrane and constrain its connections with coeval strata elsewhere in Alaska. Zircon data suggest ties among the Farewell terrane, the Kilbuck terrane of southwestern Alaska (Fig. 1), and Arctic Alaska during the Neoproterozoic. Circa 2000 Ma detrital zircon populations are a distinctive feature of early(?) Neoproterozoic (to Cambrian?) quartz-rich sandstones found widely throughout the Farewell terrane. A similar, possibly coeval quartzite occurs in the Kilbuck terrane, as do orthogneiss bodies with concordant U-Pb ages of ca. 2085–2040 Ma (Bradley et al., 2014); such bodies are a possible source for at least some of the Paleoproterozoic grains in the Farewell terrane samples. Coeval Paleoproterozoic igneous rocks, or sedimentary strata with abundant detrital zircons of this age, have not been found elsewhere in Alaska, with a single, intriguing exception. The Carboniferous Nuka Formation, which crops out widely but sparsely through the Brooks Range in Arctic Alaska, yields detrital zircon populations that consist mainly of ca. 2100–2000 Ma grains (Moore et al., 1997; Bradley et al., 2014). The Farewell, Kilbuck, and Arctic Alaska terranes all experienced felsic magmatism during the early Neoproterozoic (Tonian), documented at ca. 850 Ma in all three areas and at ca. 970 Ma in the Farewell and Arctic Alaska terranes (Bradley et al., 2014).

Late Neoproterozoic rocks suggest additional connections between the Farewell terrane and Arctic Alaska. Igneous sources for Cryogenian- and Ediacaran-age grains (ca. 750–550 Ma) are not recognized in the Farewell terrane but are present in the Arctic Alaska–Chukotka composite terrane on Seward Peninsula. Late Neoproterozoic orthogneisses (687–663 Ma; 565–562 Ma) are widely but sparsely distributed in and adjacent to the Nome Complex; their ages roughly coincide with detrital zircon ages in the Ordovician rift basin deposits of the Nome Complex (Fig. 16L; Till et al., 2014a) and in the Windy Fork and Lone Formations in the Farewell terrane (Figs. 16R and 16S). Unit O<t in the York Mountains on the Seward Peninsula is in part coeval with the Windy Fork and Lone Formations and yields a similar array of Neoproterozoic zircons (Fig. 16K; Fig. S1 [footnote 1]; Amato et al., 2009; Dumoulin et al., 2014a), although it contains more Mesoproterozoic grains.

Paleogeographic reconstructions for the late Neoproterozoic and early Paleozoic show a long-lived association among northern Laurentia, Siberia, and Baltica after the breakup of Rodinia (e.g., Li et al., 2008). The few speculative reconstructions that include displaced Arctic terranes discussed herein typically show the Farewell terrane more closely associated with Siberia through much of the Paleozoic (Colpron and Nelson, 2011; Beranek et al., 2013a). The Arctic Alaska–Chukotka and Alexander terranes are more commonly shown in association with the Timanide orogen, a late Neoproterozoic belt that formed along the eastern margin of Baltica (e.g., Gee and Pease, 2004; Amato et al., 2009; Dumoulin et al., 2012; Beranek et al., 2013a; Ayuso and Till, 2014; Till et al., 2014a).

Detrital zircon ages of ca. 750–550 Ma from the Lone Mountain area are a reasonable match for both igneous and detrital zircon ages from rocks involved in the Timanide orogeny (Fig. 2) and with known sources in Arctic Alaska as described here. Thus, we speculate that the Farewell terrane might have developed within or more proximal to the Timanide orogen together with Arctic Alaska during the late Neoproterozoic and perhaps into early Paleozoic time. In Figure 17A, we show an alternative position for the Farewell terrane in the schematic Cambrian paleogeographic reconstruction from Beranek et al. (2013a) that permits earlier ties to Siberia but also emphasizes geologic connections with terranes more strongly associated with eastern Baltica.


Faunal affinities, combined with U-Pb zircon data, indicate that ties between the Farewell terrane and Arctic Alaska continued into the early Paleozoic. Links between the Alexander terrane and the Farewell terrane may have been initiated during this time. Robust paleontologic data support proximity of the Farewell terrane and Arctic Alaska during the Ordovician; a distinctive biota that includes Siberian, Laurentian, and lesser Baltic endemics occurs in both regions (Dumoulin et al., 2002, 2014a; Blodgett, et al., 2002; Rasmussen et al., 2012). However, few paleogeographically distinctive Ordovician fossils have been reported from the Alexander terrane. Possible ties between the Alexander terrane and the Seward Peninsula during this time are suggested by co-occurrence in both areas of the coral Reushia sp., known elsewhere only in the Altai Mountains, and the gastropod Daidia sp., a Laurentian endemic (Rohr et al., 2013, 2014; Dumoulin et al., 2014a). Conodonts with Siberian affinities have been found in the Alexander terrane (Dumoulin and Harris, 2012), but the Laurentian conodonts that occur with them in the Farewell terrane and Arctic Alaska (Dumoulin et al., 2002, 2014a) have not.

Detrital zircon data suggest that all three terranes may have been near each other during at least some parts of the Ordovician. Ordovician strata in the Farewell terrane (Post River Formation), Arctic Alaska–Chukotka (units Oal and O<t, Nome Complex), and the Alexander terrane (Donjek assemblage, Field Creek volcanics, Banks Island assemblage [part], Descon Formation, Moira group) all have detrital zircon age spectra dominated by similar late Neoproterozoic and/or Early Ordovician probability peaks (Fig. 16; Fig. S1 [footnote 1]; Dumoulin et al., 2014a). Hf isotope values of detrital zircons in Ordovician samples from the northern Alexander terrane and the Banks Island assemblage have a mix of positive and negative values (Beranek et al., 2013a, 2013b; Tochilin et al., 2014). Hf data are not available from Ordovician rocks on Seward Peninsula or in the Farewell terrane, but Ordovician and older grains from Silurian strata in the Dillinger subterrane have a range of Hf values (B. Hampton, 2012, written commun.; Koroleski, 2014) similar to those seen in the northern Alexander terrane and Banks Island assemblage samples.

As noted earlier herein, felsic igneous rocks with ages that overlap the late Neoproterozoic peaks in the Farewell terrane detrital zircon samples occur on Seward Peninsula, and latest Neoproterozoic felsic igneous rocks are also found in the southern Alexander terrane (Tochilin et al., 2014). Local igneous sources for late Cambrian to Early Ordovician zircons are not known in the Farewell terrane or Arctic Alaska, but they do occur in the Alexander terrane (Beranek et al., 2013a; Tochilin et al., 2014; White et al., 2015). Thus, late Neoproterozoic zircons in Farewell terrane strata may have come from Arctic Alaska, Alexander terrane, and/or other elements of the Timanide orogeny. The Alexander terrane is a potential source for the late Cambrian–Early Ordovician zircons present in the Farewell terrane, Livengood terrane, and the York Mountains (unit Oal). In the Ordovician paleogeographic reconstruction of Beranek et al. (2013b), Baltica and associated terranes are shown in positions more proximal to Siberia and northern Laurentia. This is consistent with Ordovician fossil data linking the Farewell terrane and Arctic Alaska to all three of these cratons. However, we prefer a position for the Farewell terrane that is closer to Arctic Alaska and the Alexander terrane (solid outline in Fig. 17B) than to Siberia (dashed outline in Fig. 17B) in order to account for both the paleontologic and detrital zircon data described herein.


Faunal, lithologic, and detrital zircon data suggest that connections between the Farewell and Alexander terranes continued into the Silurian; ties to Arctic Alaska during this period are more tenuous. By Devonian time, the Farewell terrane’s history appears to have diverged from that of both Arctic Alaska and the Alexander terrane, although some evidence suggests that links between the Farewell and Alexander terranes continued into (or recurred during) the later Paleozoic.

Distinctive sponge-microbial reefs of late Silurian age occur in both the Farewell and Alexander terranes, as well as in the Ural Mountains of northern Baltica and the peri-Siberian terrane of Salair (e.g., Blodgett et al., 2002; Antoshkina and Soja, 2006, 2016; Soja, 2008). Similar reefs have not been recognized in Arctic Alaska, but carbonate strata of this age in this terrane are largely dolomitized and/or metamorphosed and do not preserve fine lithologic detail. Silurian fossils identified to the species or genus level in Arctic Alaska are predominantly cosmopolitan (Dumoulin et al., 2002, 2014a) and thus do not support or refute a connection with the Farewell terrane at this time.

Lithofacies patterns imply possible ties between the Farewell terrane and Arctic Alaska during the Silurian. Thick, dominantly calcareous mass-flow deposits are found in the northern and southern Seward Peninsula and the western Brooks Range; like the Terra Cotta Mountains Sandstone in the Farewell terrane, these strata began to accumulate in the mid-Silurian (Wenlock), and the synchronicity of these sequences suggests a common tectonic cause (Dumoulin et al., 2012, 2014a). Detrital zircon data that could be compared to those from Farewell terrane strata are not available from the Wenlock mass-flow deposits in Arctic Alaska, but a compositionally similar, slightly older unit in northwestern Arctic Alaska has been sampled. The upper Iviagik Group consists of mixed siliciclastic-carbonate turbidites that contain early Silurian (late Llandovery) conodonts (Dumoulin et al., 2014a); it produced a detrital zircon assemblage intermediate between that of the Post River Formation and the Terra Cotta Mountains Sandstone, with abundant ca. 700–440 Ma grains plus an array of older, mainly Mesoproterozoic, zircons (T. Moore, 2012, written commun.).

Existing data do not indicate a definitive single source for the detrital zircon characteristics of the Terra Cotta Mountains Sandstone, and thus the precise nature of any Silurian tectonic event involving the Farewell terrane remains speculative. Known igneous sources of Silurian age in the Farewell terrane are limited to a ca. 433 Ma tuff within turbidites that are correlative with the Terra Cotta Mountains Sandstone and crop out in the Mount McKinley quadrangle (Fig. 1; Bradley et al., 2018). Igneous rocks of this age are absent on the Seward Peninsula and throughout the Arctic Alaska–Chukotka terrane (Till et al., 2014a). The Alexander terrane contains ca. 500–430 Ma magmatic rocks (Gehrels and Saleeby, 1987; Cecil et al., 2011) and is thus a potential source for detrital zircons of this age. However, epsilon Hf isotope values from these magmatic rocks are positive (Cecil et al., 2011; White et al., 2015), whereas Silurian detrital zircon grains in Dillinger subterrane samples have both positive and negative values (B. Hampton, 2012, written commun.; Koroleski, 2014). Thus, the Alexander terrane alone does not seem to be a suitable first-order source for the full array of Silurian grains found in Dillinger subterrane samples. Terra Cotta Mountains Sandstone sedimentation coincided with the main Scandic–Salinian phase of the Caledonian orogeny (440–400 Ma), during which Baltica collided with Laurentia (Beranek et al., 2013b, and references therein). We interpret this correlation as evidence for a position near this collision zone for the Farewell terrane during the Silurian (Koroleski, 2014), rather than adjacent to Siberia (e.g., Colpron and Nelson, 2011), and for a linkage between generation of Terra Cotta Mountains Sandstone flysch and Caledonian orogenic processes. We show this alternative Silurian position in Figure 17C (adapted from Colpron and Nelson, 2011).

The provenance of the abundant Mesoproterozoic grains found in Dillinger subterrane samples also remains uncertain, because no Mesoproterozoic igneous rocks are known in the Farewell or Alexander terranes or Arctic Alaska. However, Mesoproterozoic detrital zircon grains are locally abundant in parts of Baltica (Fig. 2), including Silurian–Devonian strata of Novaya Zemlya (Lorenz et al., 2013), lower and middle Paleozoic units in northern Baltica and the Polar Urals (Miller et al., 2011), and late Proterozoic rocks of the Polar Urals (Andreichev et al., 2013) and elsewhere in the “Grenville-Sveconorwegian orogen” (Lorenz et al., 2012). Mesoproterozoic detrital zircons are also abundant in strata deposited along the Laurentian side of the Caledonian plate boundary, e.g., Silurian flysch sequences that drown the Franklinian passive margin in northern Canada (Beranek et al., 2015). These Baltic and northern Laurentian successions are possible sources of the Mesoproterozoic grains in the Farewell terrane.

Faunal affinities between the Farewell and Alexander terranes may continue into the Early and Middle Devonian (Emsian and Eifelian; Blodgett et al., 2002), and detrital zircon spectra from Devonian strata in the Farewell and Alexander terranes and Arctic Alaska have similarities (Fig. 16). Early Devonian plutons in the Alexander terrane have U-Pb zircon ages of 414–394 Ma (Cecil et al., 2011) and are potential sources for the Early Devonian zircons in our samples from the Barren Ridge Limestone. However, lithologic data indicate a Devonian history for the Farewell terrane that differs in significant ways from that of both the Alexander terrane and Arctic Alaska. No evidence has been found in the Farewell terrane for the late Silurian–Early Devonian Klakas orogeny, which affected the southern Alexander terrane, or for thick, Lower Devonian, arkosic red beds like the Karheen Formation (Gehrels and Saleeby, 1987; Bazard et al., 1995). Quartz-rich, fluvial to shallow-marine strata of the Upper Devonian–Lower Mississippian Endicott Group, which are thick and widely distributed in Arctic Alaska (e.g., Moore and Nilsen, 1984), also have no equivalent in the Farewell terrane. Instead, shallow-water carbonate rocks and limestone turbidites make up most of the Devonian section in the Farewell terrane, along with subordinate heterolithic sandstones. Upper Devonian strata in the Mystic subterrane have features indicating a high-productivity setting, such as phosphatic nodules and sedimentary barite deposits (Reed and Nelson, 1980; Bundtzen and Gilbert, 1991; Dumoulin et al., 2014b); these distinctive deposits have no counterparts in the Alexander terrane or Arctic Alaska.

However, detrital zircon data from Carboniferous–Permian strata of the Mystic subterrane have been interpreted to indicate that linkages between the Farewell and Alexander terranes (as well as the Wrangellia terrane) existed during the late Paleozoic (Malkowski and Hampton, 2014). These data suggest that the Farewell and Alexander terranes could have remained in proximity throughout the Devonian. Additional detrital zircon data from middle to late Paleozoic strata of the Mystic subterrane might help to illuminate our understanding of the Farewell terrane’s paleogeographic position through the later Paleozoic.

Neoproterozoic through Devonian successions in the Farewell, central Arctic Alaska–Chukotka, and Alexander terranes show a similar up-section shift in detrital zircon spectra, with late Neoproterozoic grains dominant in most Ordovician and older samples, and Silurian and Mesoproterozoic grains prominent in Silurian and younger rocks. This congruent evolution supports paleontologic and lithologic evidence for a shared early Paleozoic history. Connections among the three terranes were strongest during the Ordovician and had weakened by the Devonian. Late Neoproterozoic zircons in Farewell terrane strata likely originated in the Timanide orogen along the Baltic margin. The provenance of Silurian and Mesoproterozoic zircons in Farewell terrane units is less certain, but origins in the Caledonide orogen are possible. Detrital zircon spectra from the Livengood and White Mountains terranes are consistent with previously reported links to the Farewell terrane during the early Paleozoic.

We thank the Mineral Resources Program, U.S. Geological Survey (USGS), for financial support of this study; thanks also go to Stan Finney (University of California, Long Beach) for graptolite identifications, to members of the USGS Western Alaska Range Project for useful discussions, and to reviewers Luke Beranek, Matt Malkowski, and Joel Saylor, as well as special volume editor Jeff Benowitz, for constructive suggestions on earlier versions of the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

1Supplemental Files. Description of analytical methods (text document). Figure S1: Detrital zircon age probability diagrams from units Oal and O<t, York Mountains, and Nome Complex, Arctic Alaska–Chukotka composite terrane; see Table S5 for sample descriptions and locations. Table S1: Detrital zircon U-Pb data and plots. Table S2: Detrital zircon age summary of samples from the Nixon Fork and Dillinger subterranes of the Farewell terrane, the Livengood terrane, and the White Mountains terrane. Table S3: Statistical comparisons of all detrital zircon data. Table S4: Graptolite data from the Post River Formation. Table S5: Locations and descriptions of samples analyzed from the Arctic Alaska–Chukotka composite terrane. Please visit or the full-text article on to view the Supplemental Files.
Science Editor: Raymond M. Russo
Guest Associate Editor: Jeff Apple Benowitz
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.

Supplementary data